A framework for global river flood risk assessments

HESSD9, 9611–9659, 2012

A framework forglobal river flood risk

assessments

H. C. Winsemius et al.

Title Page

Abstract Introduction

Conclusions References

Tables Figures

J I

J I

Back Close

Full Screen / Esc

Printer-friendly Version

Interactive Discussion

Discussion

Paper

|D

iscussionP

aper|

Discussion

Paper

|D

iscussionP

aper|

Hydrol. Earth Syst. Sci. Discuss., 9, 9611–9659, 2012www.hydrol-earth-syst-sci-discuss.net/9/9611/2012/doi:10.5194/hessd-9-9611-2012© Author(s) 2012. CC Attribution 3.0 License.

Hydrology andEarth System

SciencesDiscussions

This discussion paper is/has been under review for the journal Hydrology and Earth SystemSciences (HESS). Please refer to the corresponding final paper in HESS if available.

A framework for global river floodrisk assessmentsH. C. Winsemius1, L. P. H. Van Beek2, B. Jongman4,5, P. J. Ward4,5, andA. Bouwman3

1Deltares, P.O. Box 177, 2600 MH Delft, The Netherlands2Department of Physical Geography, Faculty of Geosciences, Utrecht University,Utrecht, The Netherlands3PBL, The Netherlands Environmental Assessment Agency, The Netherlands4Institute for Environmental Studies, VU University Amsterdam, The Netherlands5Amsterdam Global Change Institute, VU University Amsterdam, The Netherlands

Received: 15 August 2012 – Accepted: 16 August 2012 – Published: 21 August 2012

Correspondence to: H. C. Winsemius ([email protected])

Published by Copernicus Publications on behalf of the European Geosciences Union.

9611

http://www.hydrol-earth-syst-sci-discuss.net

http://www.hydrol-earth-syst-sci-discuss.net/9/9611/2012/hessd-9-9611-2012-print.pdf

http://www.hydrol-earth-syst-sci-discuss.net/9/9611/2012/hessd-9-9611-2012-discussion.html

http://creativecommons.org/licenses/by/3.0/

HESSD9, 9611–9659, 2012


assessments


Title Page



Tables Figures

J I

J I

Back Close

Full Screen / Esc



Discussion

Paper

|D

iscussionP

aper|

Discussion

Paper

|D

iscussionP

aper|

Abstract

There is an increasing need for strategic global assessments of flood risks in currentand future conditions. In this paper, we propose a framework for global flood risk as-sessment for river floods, which can be applied in current conditions, as well as infuture conditions due to climate and socio-economic changes. The framework’s goal is5

to establish flood hazard and impact estimates at a high enough resolution to allow fortheir combination into a risk estimate. The framework estimates hazard at high resolu-tion (∼ 1 km2) using global forcing datasets of the current (or in scenario mode, future)climate, a global hydrological model, a global flood routing model, and importantly, aflood extent downscaling routine. The second component of the framework combines10

hazard with flood impact models at the same resolution (e.g. damage, affected GDP,and affected population) to establish indicators for flood risk (e.g. annual expected dam-age, affected GDP, and affected population). The framework has been applied using theglobal hydrological model PCR-GLOBWB, which includes an optional global flood rout-ing model DynRout, combined with scenarios from the Integrated Model to Assess the15

Global Environment (IMAGE). We performed downscaling of the hazard probability dis-tributions to 1 km2 resolution with a new downscaling algorithm, applied on Bangladeshas a first case-study application area. We demonstrate the risk assessment approachin Bangladesh based on GDP per capita data, population, and land use maps for 2010and 2050. Validation of the hazard and damage estimates has been performed using20

the Dartmouth Flood Observatory database and damage estimates from the EM-DATdatabase and World Bank sources. We discuss and show sensitivities of the estimatedrisks with regard to the use of different climate input sets, decisions made in the down-scaling algorithm, and different approaches to establish impact models.

9612





HESSD9, 9611–9659, 2012


assessments


Title Page



Tables Figures

J I

J I

Back Close

Full Screen / Esc



Discussion

Paper

|D

iscussionP

aper|

Discussion

Paper

|D

iscussionP

aper|

1 Introduction

There is increasing attention in the scientific and policy communities for strategic globalassessments of natural disaster risks. For example, the United Nations InternationalStrategy for Disaster Risk Reduction (UNISDR) now coordinates the production of thetwo-yearly Global Assessment Report (GAR) on Disaster Risk Reduction (GAR2009,5

GAR2011) (UNISDR, 2009, 2011), which provides a global overview of risk and risk re-duction efforts, and analyses of the underlying trends and causes of risk. Furthermore,risk due to extreme events and disasters are at the core of the Managing the Risksof Extreme Events and Disasters to Advance Climate Change Adaptation (SREX) re-port of the Intergovernmental Panel on Climate Change (Field et al., 2011). Global10

risk assessments are required: by International Financing Institutes to assess whichinvestments in natural disaster risk reduction are most promising to invest in; by intra-national institutes for monitoring progress in risk reduction activities, for example thoserelated to the implementation of the Hyogo Framework for Action (UNISDR, 2005); by(re-)insurers, who need to justify their insurance coverage; and by large companies to15

assess risks of regional investments.UNISDR (2011) defines disaster risk to be a function of hazard, exposure, and vul-

nerability. Hazard refers to the hazardous phenomena itself, such as a flood event,including its characteristics and probability of occurrence; exposure refers to the lo-cation of economic assets or people in a hazard-prone area; and vulnerability refers20

to the susceptibility of those assets or people to suffer damage and loss (e.g. due tounsafe housing and living conditions, or lack of early warning procedures). Throughoutthis paper, we have used the same terminology as UNISDR (2011).

The GAR2009 and GAR2011 reports show current estimates of global risk in termsof fatalities and economic exposure for several natural disasters, as well as trends in25

disaster risk over the past few decades. Extending these global risk assessments toinclude future changes in both natural disaster frequency and intensity (for exampledue to climate change) and socioeconomic conditions are seen as a research priority

9613





HESSD9, 9611–9659, 2012


assessments


Title Page



Tables Figures

J I

J I

Back Close

Full Screen / Esc



Discussion

Paper

|D

iscussionP

aper|

Discussion

Paper

|D

iscussionP

aper|

(e.g. Field et al., 2011). Such assessments would allow societies and the previouslymentioned stakeholders to develop and consider different options for disaster risk re-duction. The results of global risk assessments may in particular be used to comparerisks from region to region in order to decide which region deserves most commitmentto the development of risk reduction measures or mitigation procedures in a changing5

future.Flood damage constitutes about a third of the economic losses inflicted by natural

hazards worldwide and floods are, together with windstorms, the most frequent nat-ural disasters (Munich Re, 2010; UNISDR, 2009). The concentrated nature of floodsmakes them predictable in an operational context such as flood forecasting, because10

forecasts may be tailored to specific known flood prone locations and a short lead timeis sufficient to act (see e.g. Carsell et al., 2004; Verkade and Werner, 2011; Weerts etal., 2011; Werner et al., 2005). At the global scale, the local character and short timescale of floods makes prediction difficult, because global data and models are gener-ally tailored to relatively coarse spatial (and to a smaller degree temporal) resolutions.15

Moreover, the impact of local scale floods is dependent on the spatial overlap betweena flooded area and the exposed assets and inhabitants in the region. The spatial vari-ability of such exposures is often large, and there are many examples where they arein fact concentrated in flood-prone regions. The coarse resolution of global hazarddata and model outputs (e.g. around 0.5 degree scale) should therefore be tailored to20

smaller scales such as 1 km before they can be meaningfully combined with exposureand vulnerability indicators.

In this paper, we propose a global flood risk assessment framework for river floods.The framework is based on global hydrological models and global impact assessmentmodels, so that future scenario flood risk may be estimated as well. The framework25

acknowledges the spatial variability in both exposure and flood hazard, under the limi-tation that global hydrological models generally have a coarse scale resolution. In short,the framework proposes a model cascade of:

9614





HESSD9, 9611–9659, 2012


assessments


Title Page



Tables Figures

J I

J I

Back Close

Full Screen / Esc



Discussion

Paper

|D

iscussionP

aper|

Discussion

Paper

|D

iscussionP

aper|

– global forcing datasets of the current (or in scenario mode, future) climate;

– a global hydrological model;

– a global flood routing model; and

– a flood downscaling model to establish probability distributions of annual floodextremes as a measure of flood hazard.5

A second component of the framework combines these hazards with modelled floodimpacts (e.g. damage, affected GDP, affected population) at the appropriate resolu-tion to establish indicators for flood risk. The framework allows for the inclusion of re-gionally variable knowledge on flood vulnerability, through the use of spatially variableimpact models. The framework itself is presented in Sect. 2. In this section we also10

demonstrate our implementation of the framework, using a selected model cascade.We present results of our implementation in Bangladesh in Sect. 3. We discuss issuessuch as sensitivities of choices in the modelling cascade, applicability, and potentialimprovements of our application in Sect. 4. Here we also describe open research ques-tions related to the choices made in the modelling chain and invite other researchers15

or modellers to contribute to these open questions. Finally a number of conclusions ofour research are drawn in Sect. 5.

2 Description of flood risk assessment framework and implementation inGLOFRIS

2.1 General risk framework20

Generally, risk is estimated as an annual expected impact (e.g. damage), being theintegral of the probabilities of non-exceedance of certain hazardous events, multipliedby the consequence of the event (see e.g. Verkade and Werner, 2011). If this is done

9615





HESSD9, 9611–9659, 2012


assessments


Title Page



Tables Figures

J I

J I

Back Close

Full Screen / Esc



Discussion

Paper

|D

iscussionP

aper|

Discussion

Paper

|D

iscussionP

aper|

on an annual basis using annual extremes of hazardous events (common in flood riskassessment), this integral can be written as:

R =

1∫p=0

Dθ (p)dp, (1)

where R is the annual expected impact (risk), D is an impact or damage model, po-tentially consisting of both direct and indirect, tangible and intangible components, as-5

sociated with an event, along with certain event characteristics such as flood levels,extents, and durations, with annual probability of non-exceedance p [1/T]. Finally, θrepresents a number of fixed in time socio-economic factors, which determine howeasily an area is affected by floods (i.e. the vulnerability). Such factors eventually de-termine the shape of D (De Bruijn, 2005). In short, Dθ combines the exposure and10

vulnerability within an area into a damage function. In flood risk estimation, ‘events’ aretypically associated with annual time scales, meaning that not more than one extremeevent along with consequences can take place within a year. The consequence Dθ ofthe event with likelihood p can be expressed in a plethora of indicators reflecting forinstance damage (see e.g. Merz et al., 2010), affected people, loss of lives (Jonkman,15

2007) and health impact (Tapsell et al., 2002). The associated vulnerability determiningfactors, θ, may be different per consequence of interest, depending on the type of con-sequence, models used to estimate them, and socio-economic circumstances such asthe level of education, poverty, insurance coverage, and measures in place to improveresilience (e.g. dikes, flood zoning, and flood early warning procedures).20

For a global river flood risk assessment, the components used in Eq. (1) need to beestimated at the global scale, at any position of interest on the earth, and at a sufficientresolution, in order to be compatible with the spatial scale at which flood events occurand have impact. Summarised, these components include: (a) the probability distribu-tion function p of flood event characteristics; (b) the maximum exposure (i.e. maximum25

value of D); and (c) the factors θ, determining vulnerability and hence the shape of D.9616





HESSD9, 9611–9659, 2012


assessments


Title Page



Tables Figures

J I

J I

Back Close

Full Screen / Esc



Discussion

Paper

|D

iscussionP

aper|

Discussion

Paper

|D

iscussionP

aper|

A crucial part of the framework is that hazard, exposure, and vulnerability should beestablished at a high enough spatial resolution to allow for their combination into a riskestimate. This is important because if inundation occurs somewhere in a large grid box(e.g. 50×50 km2, a typical scale of a global hydrological model), this inundation onlycauses damage when the exposure within this grid box occurs at the same position5

within the grid box.Figure 1 shows an overview of our suggested framework. At the top we show the

envisaged end result, being flood risk estimates at a resolution appropriate to accom-modate both the spatial variability of the flood process and the factors determining thedegree of exposure and vulnerability. Towards the bottom, we show the components10

that lead to the envisaged end-result. In these components, two major parts can bedistinguished:

– the part in which hazard is determined in current and/or future conditions. Thispart requires a global hydrological model, along with a set of meteorological inputtime series, and a global routing model, which explicitly accounts for inundation15

of river surroundings. Finally, it requires a spatial downscaling to a resolution ap-propriate for flood risk estimation; and

– the part wherein exposure is determined in current and future conditions. Thisinformation generally comes from state-of-the art high resolution maps of typicalexposure indicators such as population and GDP, combined with future projec-20

tions of these indicators. These future projections are generally much coarserand require downscaling approaches as well, to estimate flood risk.

The individual components of the framework and how these were established in thisstudy are further described in the remaining subsections. We developed an implemen-tation of the framework called “GLObal Flood Risk with IMAGE Scenarios” (GLOFRIS).25

In this application, we made a choice in the proposed model cascade, which is de-scribed along with the general framework. This choice is certainly not exclusive and

9617





HESSD9, 9611–9659, 2012


assessments


Title Page



Tables Figures

J I

J I

Back Close

Full Screen / Esc



Discussion

Paper

|D

iscussionP

aper|

Discussion

Paper

|D

iscussionP

aper|

potential sensitivities in this choice and possible alternatives are therefore discussed inSect. 4.

2.2 Estimating global flood hazard probabilities

The flood probability of non-exceedance p, in Eq. (1) needs to be derived from ei-ther observations, or from model simulations. At a global scale, the only observations5

that are likely to provide such information with large enough density and spatial res-olution are remote sensing observations (e.g. Brakenridge et al., 2003; Prigent et al.,2007). The Dartmouth Flood Observatory (DFO) is until now the only comprehensivedataset of homogeneous flood observations that could meet the objectives outlinedin this study. DFO produces flood maps at 250 m resolution over a moving window10

archive of MODIS Aqua and Terra satellite imagery. Besides a near real-time service,DFO also stores records of flood maps that have potential for use in global risk studies.The problems of current satellite observations in view of flood risk estimation are: (a)that the collected time series are generally not long enough to provide an estimateof the current flood inundation probability distribution function p; (b) that only inunda-15

tion extent (and not depth) is provided; (c) that the data do not allow a user to makefuture projections of flood risk under climatic and/or man-made change; and (d) that,although a moving average archive is used, images are potentially partly obscured byclouds, which may render hazard estimates based on the statistics of the database toopositive. Cloudy conditions are in fact to be expected during significant flood events.20

2.2.1 Global hydrology: PCR-GLOBWB

To estimate global flood exposure probabilities, we therefore propose to use a globalhydrological modelling cascade instead. Such a modelling cascade was also recentlyproposed by Pappenberger et al. (2012). These models are potentially biased, uncer-tain, and of low spatial resolution, but they do have the potential to provide long time25

series of flow and inundation conditions, which lead to an estimate of the probability

9618





HESSD9, 9611–9659, 2012


assessments


Title Page



Tables Figures

J I

J I

Back Close

Full Screen / Esc



Discussion

Paper

|D

iscussionP

aper|

Discussion

Paper

|D

iscussionP

aper|

distribution of flow characteristics, and provide consistent spatial information. Impor-tantly, they have the ability to provide future projections when forced with scenariodata.

In Table 1, we present the requirements of the model, as well as the input datasetsto the model, in order to provide the required information. The table also presents5

justifications for the presented requirements. We discuss potential alternatives to thechoices made in this study in Sect. 4.

In GLOFRIS, we have selected to use the macro-scale hydrological model PCR-GLOBWB. This model represents the terrestrial part of the global hydrological cycle bymeans of a regular grid and discrete time steps, typically with a spatial resolution of10

0.5◦ and daily temporal resolution on the global scale. More details are given in VanBeek and Bierkens (2009); Bierkens and Van Beek (2009). This model obeys to therequirements given in Table 1.

The required forcing was established using a 30-yr (1961–1990) combination of grid-ded monthly in-situ observations of the Climate Research Unit (New et al., 2002) com-15

bined with the ECMWF 40-yr re-analysis (ERA40, Uppala et al., 2005). This datasetobeys the requirements that monthly volumes are represented as much as possi-ble, while the temporal variability is included as well. The procedure to combine bothdatasets is described by Sperna Weiland et al. (2010).

2.2.2 Global hydraulics: PCR-GLOBWB dynamic routing20

To convert the specific discharges from global hydrology, a river routing is required thatincludes overbank storage. One of the first attempts at global inundation modelling wasperformed by Yamazaki et al. (2011). They established a sub-grid variable global riverrouting model, called CaMa-Flood, which describes floodplain inundation dynamicsbased on a subgrid-variable parameterisation of floodplain topography. The output of25

this routing model is water storage, water level, flooded area, and routed dischargewithin each 0.25 degree grid cell and each (daily) time step. The output inundationdynamics of such a global river routing model may be used to establish an estimation

9619





HESSD9, 9611–9659, 2012


assessments


Title Page



Tables Figures

J I

J I

Back Close

Full Screen / Esc



Discussion

Paper

|D

iscussionP

aper|

Discussion

Paper

|D

iscussionP

aper|

of global flood statistics, i.e. a probability distribution function of flood characteristics p,but do not deliver this in the required detail for a risk assessment. This requires higherresolution information on flood statistics.

In GLOFRIS we use the PCR-GLOBWB extension for dynamic routing DynRout. Itis similar to the procedure of Yamazaki et al. (2011) in that it converts the sum of spe-5

cific discharge and the direct gains and losses from PCR-GLOBWB in river discharge,as well as overland flow in flood plain areas outside the river banks. The theory andparameterisation of the DynRout extension are further described in Appendix A.

2.2.3 Flood statistics

To derive annual flood extremes within GLOFRIS, PCR-GLOBWB and DynRout have10

been run over the 30-yr period (1961–1990) using the aforementioned input set basedon CRU and ERA40. As a further demonstration of the framework in a changing cli-mate, two future conditions were derived using GLOFRIS by running the hydrologicalmodel with the ECHAM5 and HadGEM2 model outputs in the 2050 climate conditions,forced by the IPCC SRES A1B scenario. Bias correction on the monthly means has15

been applied on the 2050 time series prior to running the model. The annual floodprobability distribution of the current and future series is estimated by assuming thateach annual extreme has an equal probability of occurrence. This principle forms theprobability distribution of flood characteristics p (see Eq. 1) at the 0.5×0.5 deg. reso-lution in current and future climate. To this end we have extracted the annual maximum20

flood volume in each grid cell, resulting in 30 global maps of flood volumes per climatecondition.

2.2.4 Downscaling of floods to appropriate resolution

The major problems of using outputs from global hydrological models for flood riskestimation are:25

9620





HESSD9, 9611–9659, 2012


assessments


Title Page



Tables Figures

J I

J I

Back Close

Full Screen / Esc



Discussion

Paper

|D

iscussionP

aper|

Discussion

Paper

|D

iscussionP

aper|

– The outputs of hydrological models are biased due to errors in model inputs, inparticular rainfall, and uncertainties in the parameterisation of dominant hydrolog-ical processes (see e.g. Haddeland et al., 2011).

– There are uncertainties in the parameterisation of the river channel dimensions. Inparticular, no information on natural or man-made levees is included in such mod-5

els. Parameterisations of natural embankments and associated channel dimen-sions are generally assumed, for instance by assuming a linear scaling betweenoutlet dimensions and upstream dimensions (Decharme et al., 2008) or an empir-ical relationship with long-term runoff (Yamazaki et al., 2011). In reality, populatedareas often have a higher man-made embankment to decrease the frequency of10

flooding.

– The spatial resolution is generally not adequate for risk estimation (∼ 0.5 de-grees), as mentioned in Sect. 2.1. Within each cell, the distribution of socio-economic conditions as well as flood hazard may be large. Data on socio-economics, required to estimate D, are typically available at resolutions of15

1×1 km2 in the current conditions, making 1×1 km2 an appropriate scale forglobal flood risk assessment.

For these reasons, an approach is required that reduces the bias in runoff generationand the effect of poorly estimated river channel dimensions in populated areas, whileat the same time increasing the resolution of the results to a meaningful spatial scale.20

To this end, we propose a downscaling of the global model results to a local scale.This downscaling may then be applied on any region, for which flood risk estimates arerelevant. This could be done, for example, at the country or basin-scale. To handle theaforementioned problems, the downscaling should consist of the following steps:

– The coarse scale global modelled time series of flood volumes should be post-25

processed into annual statistics of maxima, resulting in a coarse scale proba-bility distribution of flood events. Depending on whether a dynamic or steady-state downscaling method is used, respectively, a short (e.g. monthly) time series

9621





HESSD9, 9611–9659, 2012


assessments


Title Page



Tables Figures

J I

J I

Back Close

Full Screen / Esc



Discussion

Paper

|D

iscussionP

aper|

Discussion

Paper

|D

iscussionP

aper|

around each flood event or the static annual maximum should be retrieved fromthe daily time series.

– An assumption about non-impact floods must be made. This can be done byassuming that an event flood volume with a pre-defined probability of non-exceedance pthres (or return period 1/pthres) is not impacting on the surroundings5

of the river. This assumption reflects: (a) the fact that a given volume of water (orother characteristics, important for flood risk estimation) associated with a returnperiod safety level 1/pthres is captured by the presence of man-made embank-ments or flood retention areas; or (b) the fact that a flood with the return period1/pthres does not cause damage or any other negative impact, or even can be10

beneficial to the surroundings. This is for instance the case in areas where ricecultivation relies on annual flooding (see e.g. De Bruijn, 2005). If errors in the prob-ability distribution of flood conditions p are assumed to be largely impacted by amean bias, this procedure will also remove this bias. The value for pthres is prefer-ably selected using local knowledge of flood protection measures. The reduced15

volumes associated with the probability distribution function p are calculated as:

Vlim (p) = max[V (p)− V (pthres) ,0

], (2)

where Vlim(p) is the reduced flood volume for each probability p, and Vthres(pthres) isthe threshold volume, occurring with probability pthres at which no flood impact may beexpected. This equation implies that a flood event with a probability of non-exceedance20

smaller than pthres has no negative impact on the river’s surroundings and results in adownscaled flood map of zero depth everywhere.

– The reduced volumes Vlim(p) are downscaled over the full probability distribution,using a high-resolution Digital Elevation Model (DEM), using either a static down-scaling approach (a suggestion is given below) or a more sophisticated dynamic25

inundation model of the region (see e.g. Neal et al., 2012). The minimum pre-requisite of the downscaling procedure is that it should be mass conservative. If

9622





HESSD9, 9611–9659, 2012


assessments


Title Page



Tables Figures

J I

J I

Back Close

Full Screen / Esc



Discussion

Paper

|D

iscussionP

aper|

Discussion

Paper

|D

iscussionP

aper|

the preferred downscaling method requires discharge, besides or instead of theflood volume, then the discharge should also be reduced following the above-mentioned method, assuming that a part of the discharge with probability pthresremains within the river banks.

In GLOFRIS, we apply a static downscaling on the annual extreme volumes from5

PCR-GLOBWB. As mentioned before, the framework is not exclusive to the use ofthe method described below.

The principle idea of our downscaling approach is to impose a certain water elevation(above a certain reference, e.g. mean sea level) on river cells within a 0.5 degree pixel,and evaluate which upstream connected cells have a surface elevation lower than the10

imposed elevation in the river channel. These cells then receive a water layer, equalto the water elevation minus the surface elevation of the cell under consideration. Thisprocedure is repeated with increasing water levels, until the flood volume, imposedin the upstream cells, equals the water volume, generated by the global model in its0.5×0.5 degree river cells. A cell is considered to be a ‘river cell’ (i.e. a cell that can15

contribute to fluvial flooding) when it contains a stream with a catchment area largerthan a certain user-defined threshold. Below this threshold, the stream is not consid-ered to be significant enough to cause river flooding. In the smaller regions, flooding isassumed to be more of a pluvial or flashy nature and should be estimated from otherprocesses than river routing. The downscaling procedure is illustrated in Fig. 2.20

Our downscaling procedure ensures that the computed volume of flooded water fromPCR-GLOBWB is accounted for and is therefore mass conservative. To generate aprobability distribution of flood water levels, the downscaling routine uses the discreteextreme value distribution of flooded volumes, computed as presented in Sect. 2.2.3as input. The result is 30 high resolution estimates of water level maps with given25

probability of non-exceedance (return period), representing the fluvial flood hazard atan appropriate resolution.

9623





HESSD9, 9611–9659, 2012


assessments


Title Page



Tables Figures

J I

J I

Back Close

Full Screen / Esc



Discussion

Paper

|D

iscussionP

aper|

Discussion

Paper

|D

iscussionP

aper|

2.3 Estimation of exposure and vulnerability indicators

2.3.1 Introduction of methods

In the framework, flood risk is defined as a product of hazard, exposure, and vulnera-bility (UNISDR, 2009). The hazard is represented by the hydrological model cascade(e.g. within GLOFRIS) in the appearance of flood extent and depth. In this section we5

will discuss the possible incorporation of exposure and vulnerability indicators in theframework.

Exposure to a flood event can be subdivided into physical exposure, defined as thenumber of people and assets affected by the event, and the resulting economic ex-posure (Peduzzi et al., 2009). Economic exposure is represented by the total value10

of assets in the affected area, which can be estimated using several methodologies.Existing local and regional methodologies calculate exposure in terms of asset valuesor maximum damage values per individual property or square metre of specified landuse (Merz et al., 2010; Messner et al., 2007; Smith, 1994). These asset values aredetermined using detailed empirical damage data from past flood events or analysis of15

synthetic (what-if) scenarios (e.g. Green et al., 2011).Because detailed spatial data are not available on a global scale, exposure indicators

for global impact assessment are inevitably more generalised. Since asset values aredirectly related to GDP per capita (Green, 2010), a combination of population densityand GDP data can be used as an indicator for the total value of assets (Peduzzi et20

al., 2009). An important limitation to this approach is that GDP per capita is an aver-age of total national income, while the spatial differences are substantial (Hill, 2000).An alternative approach is the upscaling of existing regional approaches. Jongman etal. (2012a) have achieved this by taking asset values per square metre calculated forthe Netherlands; calculating the values for other countries on the basis of the relative25

GDP per capita difference; and applying the resulting square metre figures to a globalurban density map. A limitation of this approach is that the regional model is designedon the basis of a high-resolution land use map with various categories, while the global

9624





HESSD9, 9611–9659, 2012


assessments


Title Page



Tables Figures

J I

J I

Back Close

Full Screen / Esc



Discussion

Paper

|D

iscussionP

aper|

Discussion

Paper

|D

iscussionP

aper|

data is relatively coarse and has only one “urban” class. This results in a large degreeof generalisation.

Vulnerability defines to what extent the exposed people and assets are adverselyaffected by the flood event (Jha et al., 2012; UNISDR, 2011). With increasing vulner-ability, the exposed assets will be affected to a larger degree. The accepted standard5

method for including local vulnerability in land use based flood risk assessment is theuse of depth-damage curves (Green et al., 2011). Depth-damage curves are math-ematical functions representing the percentage loss of the total asset value with in-creasing inundation depth. On a societal scale, important vulnerability factors that areknown to influence flood impact are corruption, equality, bureaucratic quality, law and10

order, ethnic and religious tensions, government stability, and democratic accountabil-ity (Ferreira et al., 2011). Estimations of these social and governance aspects, in termsof vulnerability indicators, are available on country-level for the entire world (e.g. Kauf-mann et al., 2011; World Bank, 2012).

In GLOFRIS, we used two methodologies for the quantification of flood impact.15

The first method is based on population density and GDP per capita estimates(Sect. 2.3.2) and the second method on urban density and maximum damage esti-mates (Sect. 2.3.3). The first will be referred to as the population method in the remain-der of this paper. The latter is referred to as the land use method. Both methodologiesuse high-resolution (1x1 km) data of population, GDP per capita, and urban density20

distribution for current impact assessment. Projections up to 2050 are conducted us-ing 0.5 degree resolution GDP and population estimates from the Integrated Model toAssess the Global Environment (IMAGE, Bouwman et al., 2006) and the Global Inte-grated Sustainability Model (GISMO, PBL, 2008). Population and GDP were derivedfrom GISMO. IMAGE provides scenarios of world-region resolution population. GISMO25

is a spatially more explicit scenario model, and ingests IMAGE population scenariosand downscales these to a 0.5 by 0.5 degrees spatial level based on distinctions be-tween urban and rural areas and is provided in the current condition and different future

9625





HESSD9, 9611–9659, 2012


assessments


Title Page



Tables Figures

J I

J I

Back Close

Full Screen / Esc



Discussion

Paper

|D

iscussionP

aper|

Discussion

Paper

|D

iscussionP

aper|

scenarios. More details on the GISMO approach to population downscaling may befound in Van Vuuren et al. (2007).

2.3.2 Method 1: Population-scaled GDP (population method)

Maximum exposure: the Gross Domestic Product – Purchasing Power Parity(GDP/PPP) per capita is used as an approximation of assets. Country averaged5

GDP/PPP from the World Bank’s world development indicators (WDI) were used forthe base year and divided over the 0.5 degree population map, assuming each indi-vidual would be equally rich. Economic growth rates from the IPCC-SRES scenarios(IPCC, 2000) were used as future scenario. Regional economic growth rates combinedwith the GDP per country in the base year results in GDP for the scenarios per 0.510

degree pixel.The future 0.5 degree maps of GDP per capita were further downscaled to 1×1 km2

using a detailed population map of the current situation, which is projected into thefuture. Three datasets make the downscaling possible: (1) the Landscan 2007 popula-tion dataset, which counts the population on a spatial scale of 30 by 30 arc seconds15

(∼ 1×1 km2) for the whole world, (2) the CIESIN GPW3 GRUMP urban and rural extentdataset which distinguish urban and rural land use; and (3) GLOBCOVER 2006, theGlobal Land cover database which is used to prevent allocation of population in bareareas. The Landscan data on population is chosen instead of the Gridded Population ofthe World (GPW) on which the GRUMP dataset is based because of detailed resolution20

and advanced modelling practices (Meijer et al., 2006).The downscaling is performed by assigning urban and rural population 1×1 km2

cells within each 0.5 degree cell. These are assigned different population densities.The Landscan 2007 population data is assigned to be urban or rural by using theGRUMP urban or rural extent. For each 1×1 km cell within a 0.5 degrees cell, the25

fraction of urban and rural population is calculated using the GRUMP and Landscan2007 data. The GISMO urban and rural population at the 0.5 degrees scale were dis-tributed over the 1×1 km cells using the calculated fraction. Cells with urban population

9626





HESSD9, 9611–9659, 2012


assessments


Title Page



Tables Figures

J I

J I

Back Close

Full Screen / Esc



Discussion

Paper

|D

iscussionP

aper|

Discussion

Paper

|D

iscussionP

aper|

according to GISMO, but without urban extent according to GRUMP in the current situa-tion, are assigned the summed urban and rural population. In case Landscan indicatesno population, but GISMO does, the population is equally distributed over a cell. Incase GLOBCOVER 2006 indicates bare areas, water bodies, or permanent snow andice, no population is assigned.5

Vulnerability : the most simple DDF shape was used. Affected GDP was assumed toincrease linearly with water level from a damage of zero for a water level of zero, to amaximum affected GDP (see above) at a level of 3 m.

2.3.3 Method 2: Land use based damage (land use method)

Maximum Exposure: in most flood damage modelling studies, economic exposure is10

based on a discrete land cover map, whereby each land cover class has a correspond-ing asset value. In this study, we used fractional land cover maps, which show thefraction of land within each grid cell covered by each land cover class. Since the aim ofthe paper is to demonstrate the potential application of flood risk modelling techniqueson a global scale, we used a simple classification of land cover into three classes: high-15

density urban; low-density urban; and non-urban. For the reference time-period (2010),we firstly used land cover data from two sources: MODIS (Schneider et al., 2009); andthe GRUMP dataset (CIESIN and CIAT, 2009) to create a discrete land cover mapshowing urban and peri-urban areas, using the approach of Kummu et al. (2011).Secondly, we created a set of three fractional land cover maps (high-density urban,20

low-density urban, and non-urban). For the high-density urban map, we assumed afractional cover of 0.75 in those cells classed as urban in the discrete land cover map(since MODIS assigns cells as urban where 50 % or more of the cell is urbanised), anda fractional cover of 0 in all other cells; and then applied a linear interpolation methodwith a maximum distance of 5 km to derive a high-density urban area map with frac-25

tional values between 0 and 0.75. For the low-density urban area map, we assumedthe fractional cover to be 0.25 for those cells classed as peri-urban in the discrete landcover map (except where this led to a total area fraction greater than 1 when added

9627





HESSD9, 9611–9659, 2012


assessments


Title Page



Tables Figures

J I

J I

Back Close

Full Screen / Esc



Discussion

Paper

|D

iscussionP

aper|

Discussion

Paper

|D

iscussionP

aper|

to the urban map). The non-urban land use map was derived by subtracting both theurban and peri-urban maps from unity. For 2050, we projected the change in the per-centage of high-density and low-density urban area per grid cell, based on the methoddeveloped and validated in Jongman et al. (2012a). The projection method is shown inequation 1, whereby Aurban [L2] is the urban area per grid cell, ρ [-] is the total popula-5

tion of the country, υ [-] is the fraction of the population living in urban settlements (UN,2010), and t is the specific year of analysis.

Aurban (t) = Aurban (t−1)ρ (t)υ (t)

ρ (t−1)υ (t−1)(3)

The projection has been validated for the period 1970–2005 on the basis of HYDEhistoric urban density data (Klein Goldewijk et al., 2011). We found that the projected10

density using Eq. (3) has an overall R2 of 0.99 with urban density from the HYDE data,and is significant at a 99 % confidence interval. Note that we did not allow for an expan-sion in urban area, since such an assessment would require sophisticated land covermodelling at the global scale. Finally, we assigned an economic asset value to eachland cover class, using the method applied in Jongman et al. (2012a). As a basis for15

asset values we use the figures calculated for the Netherlands in the Damage Scan-ner model (Klijn et al., 2007), which we adjust for each country relative to its GDP percapita at Purchasing Power Parity. For Bangladesh, the resulting asset values for 2010and 2050 respectively are: (a) $ 33 m2 and $ 177 m2 for urban area; and (b) $ 10 m2

and 53 m2 for peri-urban area.20

Vulnerability : here, we use DDFs derived from the Damage Scanner (Aerts et al.,2008; Klijn et al., 2007) as a demonstration of methods. These DDFs were based onDutch data. We apply a different DDF for urban and for peri-urban areas. Althoughmore advanced than our first linear DDF, it should be noted that the identification ofimproved international vulnerability assessment methods should be a research priority25

(Jongman et al., 2012b).

9628





HESSD9, 9611–9659, 2012


assessments


Title Page



Tables Figures

J I

J I

Back Close

Full Screen / Esc



Discussion

Paper

|D

iscussionP

aper|

Discussion

Paper

|D

iscussionP

aper|

2.4 Risk

Flood risk has been estimated in terms of annual expected values of affected GDP, anddamage. This is done by integrating Eq. (1) using the established probability distributionof flood hazard p with associated flood levels, described in Sect. 2.2, and the damagemodels D, which reflect exposure and vulnerability θ due to local asset characteristics.5

3 Results

The steps described in the previous section lead to:

– a probability distribution of river flood volumes in all grid cells of the global hy-drological model PCR-GLOBWB and DynRout. The model setup is described inSects. 2.2.1 and 2.2.2, whereas the derivation of the probability distribution is10

given in Sect. 2.2.3.

– localised probability distributions of river flood levels, taking into account the localdischarge capacity of rivers (method described in Sect. 2.2.4).

– localised flood risk estimates expressed as annual expected values of a certainflood impact. In this case alternative impacts are demonstrated being: the ex-15

pected value of annual affected GDP, based on the population method (all equallyrich); and the expected value of annual damage, established with the land usemethod, which focuses more on the actual asset value (described in Sect. 2.3).

To demonstrate the framework’s results, we have established the downscaling and riskestimates for Bangladesh as a first application.20

3.1 Flood hazard estimation and validation

The flood maps, of which an example is shown in the left hand-side in Fig. 3, areproduced under the assumption that a 2-yr flood is within the river’s drainage capacity.

9629





HESSD9, 9611–9659, 2012


assessments


Title Page



Tables Figures

J I

J I

Back Close

Full Screen / Esc



Discussion

Paper

|D

iscussionP

aper|

Discussion

Paper

|D

iscussionP

aper|

Furthermore, river flooding is considered to play a role only in streams with a Strahlerstream order of 6 and larger (Strahler, 1964). The sensitivity of these assumptions isfurther investigated in our discussion. We have performed a rough validation of themodelled flood hazard against the flood extent maps of Dartmouth Flood Observatory(DFO, http://floodobservatory.colorado.edu/). This validation is limited as the modelled5

once in 30 yr flood is not per se fully equivalent with the DFO observed flood extent.In fact, the return periods belonging to the flood map may be different for each pixel.Furthermore, this validation does not give any information about the correctness of thefrequency of occurrence of flood levels above a certain depth from our model cascade,which would require a long time series of annual extreme flood levels. DFO does not10

show any flood levels but merely a classification of flooded and non-flooded areas.Finally, DFO also shows floods due to impeded drainage and coastal floods, processeswhich are not modelled by the GLOFRIS model cascade yet.

Figure 4 shows our 30-yr flood extent (left) as well as the maximum documentedDFO flood extents over South-East Asia. Flood extent around the large rivers such as15

the Ganges Brahmaputra, the Chao Phraya and the Indus are quite well estimated. Itis evident that areas with probably relatively shallow inundation, further away from themain river are less well simulated. Our inundation algorithm is based on the principlethat floods are generated by backwater from large rivers. In the less well representedareas, the inundation could be caused by a different phenomenon such as overland20

flow or flooding by local rainfall. To improve the results in these areas, a more dynamicdownscaling approach could be used, e.g. suggested by Neal et al. (2012). Coastaldeltaic areas such as the lower Mekong and Ganges Brahmaputra are also likely to beunderestimated by GLOFRIS, because coastal flooding plays a dominant role in theseareas.25

Figure 5 shows a more zoomed extent over Bangladesh. Both figures reveal a rea-sonable resemblance in flood patterns. In the Bangladesh case, it seems that in theSouth-East of the country, flood extent is somewhat overestimated while in the North-East there is some underestimation. Probably, the drainage network in the South-East,

9630





http://floodobservatory.colorado.edu/

HESSD9, 9611–9659, 2012


assessments


Title Page



Tables Figures

J I

J I

Back Close

Full Screen / Esc



Discussion

Paper

|D

iscussionP

aper|

Discussion

Paper

|D

iscussionP

aper|

estimated by the model, is connected to the outflowing reaches, while in reality thisconnection does not exist. The North-East region belongs to the HOAR sub-basin andis a flash flood prone region. As flash floods are not simulated by our model cascade,this could be a logical explanation for the underestimation here.

3.2 Risk estimation and validation5

The middle and right-hand plates in Fig. 3 show the resulting expected values of dam-age, based on the population method and the land use method. The spatial patterns ofthe risk estimates also reveal the underlying spatial heterogeneity of the risk-causingprocesses, being on the one hand the flood hazard, and on the other hand the distri-bution of GDP, land use, and population. This emphasises the importance of solving10

risk at an appropriate spatial scale. Naturally, the value of assets may be very inde-pendent of the population size. This explains most of the differences between the twoapproaches to estimate flood risk.

In Table 2 we summarise the expected annual damage (based on the populationmethod and land use method) for the Bangladesh case study for the different haz-15

ard and exposure scenarios. In current conditions, the expected annual damage isca. $ 740 million according to the land use method, while it is $ 2183 million using thepopulation method. Under the future scenarios of hazard and exposure change, thesevalues increase by a factor 22–30 and 21–28 respectively, depending on the GCMused. We see that the effects of simulated change in exposure only (increase by a20

factor 11 and 7 respectively) are much higher than those of climate change only (in-crease by factor 3 to 4, depending on GCM and impact assessment method used).The results for the two impacts assessment methods differ by a factor of approximately3, while the relative differences from current to future conditions are approximately thesame for the two methods. The most extreme once in 30-yr damage in our calculations25

is approximately $ 4500 million.Furthermore, we have performed a limited validation of the risk results against ob-

served damages cited in existing literature and databases. Major river floods took place9631





HESSD9, 9611–9659, 2012


assessments


Title Page



Tables Figures

J I

J I

Back Close

Full Screen / Esc



Discussion

Paper

|D

iscussionP

aper|

Discussion

Paper

|D

iscussionP

aper|

in Bangladesh in 1998 and 2004. Estimates of the damages as a result of these floodscan be found in the EM-DAT database (EM-DAT: the OFDA/CRED International Disas-ter Database – www.emdat.net – Universite catholique de Louvain – Brussels – Bel-gium) and in the report of World Bank (2010). These values, adjusted to $ PPP2010values can be found in Table 3. The flood volumes of 1998 are estimated to have had5

a return period of 52 yr according to the World Bank (2010) and of ca. 60–70 yr onthe Brahmaputra and 30 yr on the Ganges according to Monirul Qader Mirza (2003).Hence, in Table 3 we compare our results for the simulated 30 yr return period with theobserved results for 1998, noting that the actual return period in 1998 was in fact higher.For 2004, the World Bank (2010) estimated the flood volume to have a return period of10

20 yr. The closest available simulated return period in our study is 15 yr, and hence inTable 3 we show these figures for the simulated values in 2004. The simulated damage(land use approach) is of the same order of magnitude as the observed results, thoughit is somewhat lower. However, this is to be expected since: (a) the simulated returnperiod is 30 yr, whilst the observed 1998 data actually refer to a less frequent flood15

event; and (b) our model only simulates damage in urban and peri-urban areas, whilstagricultural damages are also a major component of total damages in Bangladesh. Forexample, according to World Bank (2010) about a third of the damages in 1998 were inthe agricultural sector. The population method estimate, on the other hand, somewhatoverestimates the damages. Again, this is not surprising since the reported damages20

are direct damages to assets, whilst an estimate of affected GDP implicitly entails alarger pool of indirect losses.

The results in this application show that, although the global hydrology is solved ata coarse resolution, our application of the framework can provide information that isconsistent with the spatial scale at which fluvial floods occur (∼ 1 km2).25

9632





www.emdat.net

HESSD9, 9611–9659, 2012


assessments


Title Page



Tables Figures

J I

J I

Back Close

Full Screen / Esc



Discussion

Paper

|D

iscussionP

aper|

Discussion

Paper

|D

iscussionP

aper|

4 Discussion

Our GLOFRIS implementation is just one model cascade that could be followed, andthe application of this framework with another cascade will affect the results of our riskestimates. All components of the framework could be replaced by other models andmethods. For instance, another model cascade, limited to flood hazard only, was pre-5

sented recently by Pappenberger et al. (2012). In cases in which risk is expressed as afractional change from one place to the other or one period to the other, the sensitivityto the chosen model cascade is likely to be smaller than a case where indicators areused that are directly related to physically measurable units in a certain location (e.g.damage in US$ PPP). In this case the absolute uncertainty due to model choices be-10

comes more important to address. In this discussion we focus on the impact of some ofthe choices made in the application of our framework, and discuss additional choicesthat could be varied in future research.

4.1 Climate input uncertainty

The choice was made to run PCR-GLOBWB over a 30 yr period, based on a combi-15

nation of CRU and ERA40 data. This assumes that the CRU-ERA40 data are repre-sentative for the current climate and that 30 yr is a long enough period to establishthe required extreme value probability distribution. Extreme probability distributions arenon-Gaussian of nature which makes in particular the tails of the probability distribu-tion uncertain (see e.g. Ruff and Neelin, 2012) and will become more accurate when a20

longer time series can be used. The required forcing of candidate hydrological modelsgenerally consists of global precipitation, temperature, and potential evaporation, thelatter often in turn made dependent on net incoming short-wave radiation, long-waveradiation, temperature, wind speed, and humidity. Candidate input datasets should pre-serve long-term averages as well as temporal variability. A number of potentially suit-25

able datasets exist, however all datasets may still suffer from errors due to poor sam-pling, undercatch (Biemans et al., 2009; Weedon et al., 2011), or limited representation

9633





HESSD9, 9611–9659, 2012


assessments


Title Page



Tables Figures

J I

J I

Back Close

Full Screen / Esc



Discussion

Paper

|D

iscussionP

aper|

Discussion

Paper

|D

iscussionP

aper|

of variability. How sensitive our results are to the choice of the dataset is yet to be in-vestigated. It should be kept in mind that the goal of a global flood risk analysis is toidentify factors such as spatial differences in risk and changes in risk given certain pro-jected scenarios. It is therefore important to consider that bias in the input dataset mayimpact on the absolute results, but is likely to impact comparative results from place to5

place or scenario to scenario to a lesser degree, as long as a globally consistent mod-elling cascade is used. In future work, we will investigate this sensitivity by running ourmodel with the WaterMIP dataset from the EU-WATCH project group (Weedon et al.,2011) as well as the ERAInterim-GPCP dataset (Balsamo et al., 2010). The WaterMIPset is also based on CRU and ERA40 but also includes the GPCP rainfall data to cor-10

rect the monthly accumulated rainfall. Furthermore, a longer time period is used whichwill result in a more accurate description of the extreme value probability distribution.The ERAInterim-GPCP dataset combines the Global Precipitation Climatology Project(GPCP) dataset of monthly rainfall observations, based on both in-situ and satellite ob-servations (Huffman et al., 2009) with the recent ERA-Interim re-analysis rainfall time15

series.

4.2 Hydrological model uncertainty

Although the chosen hydrological model has been rigorously validated over many re-gions using a number of data sources, many other hydrological models exist that pro-vide similar outputs as PCR-GLOBWB. In the Water Model Intercomparison Project20

(WaterMIP), considerable work has been carried out to compare and explain differ-ences between global hydrological models and land surface models (Haddeland etal., 2011). The closest to the required information for flood risk estimation is the com-parison of runoff. WaterMIP has compared the runoff from different models on annualand monthly time scales, and revealed considerable differences between all models25

considered. In this comparison, it should be noted that the models used in the Wa-terMIP project differ considerably in their goal. For instance, the Lund-Jena-Potsdammanaged Land (LPJmL) model (Bondeau et al., 2007; Rost et al., 2008) is designed

9634





HESSD9, 9611–9659, 2012


assessments


Title Page



Tables Figures

J I

J I

Back Close

Full Screen / Esc



Discussion

Paper

|D

iscussionP

aper|

Discussion

Paper

|D

iscussionP

aper|

for the purpose of vegetation type simulations. The runoff generation processes maytherefore be simplified to a degree that is sufficient for modelling the vegetation andrelated moisture stores. The runoff generation process is modelled as a saturationexcess function, which means that runoff is not gradually increasing with increasingmoisture state, but instead occurs only when full saturation is reached. For monthly ac-5

cumulated runoff values, this is not necessarily problematic, but for shorter time scalesat which flood genesis is relevant, such detail in hydrological processes is mandatoryand a non-linear function with soil moisture is required. Model comparisons such asWaterMIP are at the moment focusing on river flows at large (e.g. monthly or seasonal)time scales. However, in such comparisons, more emphasis should be paid to the re-10

production of flood genesis and the shape of the hydrograph in general, in order totailor selections of model candidates for risk assessments of extreme events. This hasso far not been considered in comparison studies of global hydrological models andis therefore still an open issue in this paper. Studying the behaviour of hydrologicalmodels during extremes may be done by using hydrograph signatures, such as auto-15

correlation, variance, and flow duration curves (see e.g. Gupta et al., 2008; Westerberget al., 2011; Winsemius et al., 2009). To address the sensitivity of the model choice, weadvocate using a multi-model ensemble, where the weight of each ensemble membercould be chosen based on performance measures, which focus on the flood genesis.We invite other research groups to join this research effort with alternative hydrological20

models.

4.3 Sensitivities in hydrological downscaling

A number of choices are also required in the downscaling procedure. Figure 6 demon-strates for Bangladesh how the once in 30-yr flood (i.e. the most extreme flood in thecomplete time series) is affected by the choice of a different threshold above which25

rivers are classified as flood-prone rivers (stream order 6 or 7), or a different returnperiod threshold for the flood levels expected not to cause any inundation or damage inthe river’s surroundings (return period of 2 or 5 yr). It is evident that these choices have

9635





HESSD9, 9611–9659, 2012


assessments


Title Page



Tables Figures

J I

J I

Back Close

Full Screen / Esc



Discussion

Paper

|D

iscussionP

aper|

Discussion

Paper

|D

iscussionP

aper|

at least moderate influence on the results. A stream threshold above 6, compared to 7,means that more river channels take part in the flooding process, and thus the watervolume from a certain 0.5 degree pixel is distributed over a larger area. An obvious ex-ample of this is an artefact in the very South of the country, where a large flood pronearea appears when stream order above 6 is selected. This area is not flood-prone when5

the threshold is set to 7. The overall result is that a larger region becomes flood-pronebut that the water levels in these areas become somewhat lower. Obviously, selectinga lower safety level also results in a larger flood hazard, although this effect is lessobvious.

These two choices should be made differently per region of interest, depending on10

the physical characteristics of the river system and the level of (natural or man-made)protection against flooding. Preferably, the choices should be made based on localknowledge. In addition, the choices may be validated by comparison with remote sens-ing information, e.g. flood maps from the Dartmouth Flood Observatory.

Other downscaling approaches can also be used. For example, Neal et al. (2012)15

have developed a large-scale applicable dynamic modelling approach, which could beused to perform a more physically based downscaling, which not only preserves themass balance but also momentum.

4.4 Limitations of impact models

In this study we have applied two separate methods for estimating flood risk on the20

basis of global hazard, exposure, and vulnerability data. One method is based on pop-ulation density and GDP per capita, and the other on land use and estimated assetvalues. Both methodologies have limitations that result from their approaches and as-sumptions.

The population method uses a linear relationship of GDP per capita to quantify the25

value of assets at risk of flooding. In reality, this relationship may be more complex.In addition, a longer time-scale average GDP/population estimate rather than a snapshot in time would be more appropriate, given that our flood risk framework focuses

9636





HESSD9, 9611–9659, 2012


assessments


Title Page



Tables Figures

J I

J I

Back Close

Full Screen / Esc



Discussion

Paper

|D

iscussionP

aper|

Discussion

Paper

|D

iscussionP

aper|

on climatological risk estimates. Furthermore, the homogenous distribution of GDPover all areas within a country entails the assumption of equal productivity by eachinhabitant. In reality, important sources for GDP are often focused on several specificsites or regions, especially in countries where a large part of the domestic productresults from natural resource extraction (Hill, 2000; Jongman et al., 2012a).5

The land use method also has the limitation that the estimated maximum damagevalues are directly linked to GDP per capita. Maximum damage values should ideallybe calculated on the basis of more information, such as family income and property val-ues. Furthermore, the land use method applies the same vulnerability (depth-damage)functions in each region, while in reality the true vulnerability will vary within and be-10

tween countries (Jongman et al., 2012b; Merz et al., 2010). More research is neededto make spatial differentiations in vulnerability at the global scale, for example on thebasis of estimated building material and quality.

Finally, the available spatial population and land use data make it difficult to produceconsistent projections of future flood risk. Existing global models of future land use are15

limited in that they operate on a low resolution (e.g. 0.5◦ or 50 km) and focus predomi-nantly on agriculture and vegetation (Verburg et al., 2011). The current understandingof urban expansion over time is limited and fragmented (Seto and Shepherd, 2009).However, ongoing research activities in this field (e.g. Letourneau et al., 2012; Seto etal., 2011) should mean that finer resolution global urban land cover models become20

available in the short-term.

4.5 Limitations

So far, we have applied our framework over a 30-yr time series, under the assumptionof a return period of non-flooding (in the Bangladesh case, this return period was seton 2 yr). This can only be done when the return period of non-flooding is consider-25

ably lower than the amount of simulated years available. In areas with high protectionstandards, the simulated time series are likely to be too short to establish a satisfyingprobability distribution of events. Therefore, the applicability of our framework is until

9637





HESSD9, 9611–9659, 2012


assessments


Title Page



Tables Figures

J I

J I

Back Close

Full Screen / Esc



Discussion

Paper

|D

iscussionP

aper|

Discussion

Paper

|D

iscussionP

aper|

now limited to areas with low protection standards. This is the case in most developingcountries. These are also the areas where our framework is most interesting. Further-more, as mentioned before, the relatively short (30 yr) simulated time series results ina relatively small amount of samples in the tail of the extreme value distribution.

Man-made interactions with the river system, such as the operation of dams and5

reservoirs, have not yet been taken into account. These could be included in futurestudy, but with the risk of incorrectly estimating the operation during flood conditions.The impact of reservoir control could result in the reduction of floods if the controllerhas proper information at hand to decide upon pre-releases, but in many cases if suchinformation is not at hand at the reservoir, they may result in larger floods if unexpected10

inflows are experienced.The effect of levee breaks is also not included, which can have large impacts on

flood patterns. For example, during the Pakistan floods in 2009 a large part of a majorembankment was destroyed by the floods, causing a completely different flood patternthan what a model would simulate. This appeals for a more interactive approach to15

mappin flood hazard, which allows for what-if scenarios on the schematisation of theelevation profile throughout a case study area. Obviously, such what-if scenarios arenot suited for a global approach such as presented here.

A further limitation so far is that whilst flood risk has been modelled as a functionof flood hazard , exposure, and vulnerability, the vulnerability has been assumed to20

remain the same in time and space. Future developments in resilience and adaptationmeasures may however reduce vulnerability (e.g. due to increased in awareness, otherbuilding methods, flood warning procedures, and so on and so forth). Again, resilienceand vulnerability are spatially diverse in nature and should be established using in-situinformation. This information may then be propagated into the damage functions. Such25

local information may be limited when considering large-scale applications.Finally, we only demonstrated our method using one climate scenario, and two GCM

models. If an in-depth investigation into the effects of climate change on flood risks isrequired, more scenarios, and in particular more GCM runs, should be considered, as

9638





HESSD9, 9611–9659, 2012


assessments


Title Page



Tables Figures

J I

J I

Back Close

Full Screen / Esc



Discussion

Paper

|D

iscussionP

aper|

Discussion

Paper

|D

iscussionP

aper|

the variability due to GCM selection is larger than the variability due to the scenariochosen (Sperna Weiland, 2011). Such an analysis is outside the scope of this paper.

5 Conclusions

An aim of our global flood risk assessment framework for river flooding is to provideflood risk maps using a globally consistent combination of models. This is done by5

producing hazard and exposure maps at an appropriate spatial scale, and combin-ing these with vulnerability relationships, in order to create risk maps. In this paper,a 1×1 km scale was used, because most of the global data used in this study (e.g.elevation, land cover, population, and urbanisation) are available at this spatial scale.

In this paper, we described a specific implementation of the framework called GLObal10

Flood Risks with IMAGE Scenarios. This implementation combines the global hydrolog-ical model PCR-GLOBWB, with a subgrid parameterised dynamic flow routing calledDynRout and a flood extent downscaling algorithm to provide a distribution functionof global flood hazard. GLOFRIS combines scenarios from the IMAGE instrument,downscaled with land use and population maps to 1×1 km, with a number of damage15

models to estimate exposure, with vulnerability in the form of depth-damage functions.From this study we may conclude that besides the global hydrological models, down-

scaling to higher resolution than the typical resolution of global hydrological models (inthe order of 50 km) is required in order to perform a feasible combination between floodhazard on the one hand, and flood exposure and vulnerability on the other hand. This20

is because the spatial variability in both hazard and potential impacts may be muchlarger than the typical resolution of global hydrological models.

To demonstrate GLOFRIS, we applied the full modelling framework on the case studyof Bangladesh. The study resulted in plausible river flood hazard maps and damageestimates, which were in the same order of magnitude as estimates from the global25

EM-DAT database and World Bank estimates. This demonstrates that the model ap-proaches used as implementation of our framework provide satisfactory results.

9639





HESSD9, 9611–9659, 2012


assessments


Title Page



Tables Figures

J I

J I

Back Close

Full Screen / Esc



Discussion

Paper

|D

iscussionP

aper|

Discussion

Paper

|D

iscussionP

aper|

Furthermore, we applied the framework using combinations of climate and socio-economic changes. Climate changes simulated by ECHAM and HadGEM A1B causeda relative increase in flood risk compared to current conditions of approximately a factor3. Socio-economic changes caused an additional relative increase of factor 10 overBangladesh. This demonstration shows that our framework can be applied to rapidly5

estimate flood risk changes over large areas and significantly attribute these changesto hazard increase or decrease (i.e. due to climate change), and changes in potentialimpacts (i.e. due to socio-economic changes).

Currently, our framework has been implemented using one cascade of models. Wehighly recommend that this approach is extended towards a multi-model approach,10

where any component in the model cascade can be exchanged with another compo-nent. This is seen as a research opportunity.

Appendix A

Description of PCR-GLOBWB dynamic routing

For those cells that are identified as stream segments, discharge is calculated from15

the kinematic wave approximation of the Saint-Venant Equation (Chow, 1988). Thecontinuity equation is:

∂Q∂x

+∂A∂t

= q, (A1)

and the momentum equation can be expressed by:

A = αQβ, (A2)20

where Q is the streamflow through the channel [L3 T−1], A is the channel cross-section[m2], q is the inflow –specific runoff- per length of channel [L2 T−1] x is the length along

9640





HESSD9, 9611–9659, 2012


assessments


Title Page



Tables Figures

J I

J I

Back Close

Full Screen / Esc



Discussion

Paper

|D

iscussionP

aper|

Discussion

Paper

|D

iscussionP

aper|

the channel [L] and t is the elapsed time [T]. q is here delivered by the water balancemodel, described above.

The coefficients a and b are obtained from Manning’s equation (Chow, 1988):

Q =R2/3√S

nA, (A3)

where R is the hydraulic radius [L], A is the cross-sectional surface [L2], S is the energy5

gradient, in the kinematic wave approximation, this gradient is assumed to be equal to

that of the river bed [-], and n is Manning’s roughness coefficient [L5/6 T−1]. R can besubstituted by A/P , where P is the wetted perimeter [L] and Eq. (22) rewritten in termsof A:

A =

[nP 2/3

√S

]3/5

Q3/5, (A4)10

which gives

α =

[nP 2/3

S1/2

]3/5

, (A5)

and

β = 3/5. (A6)

To ensure numerical stability, Eq. (A1) is applied for a variable number of sub-time15

steps per day that satisfy the Courant number everywhere.For lake and reservoir cells, which can extend over several cells to form contiguous

water bodies, discharge is calculated in analogy to the weir formula as the dischargeover a rectangular cross section in this study, although a separate scheme can be

9641





HESSD9, 9611–9659, 2012


assessments


Title Page



Tables Figures

J I

J I

Back Close

Full Screen / Esc



Discussion

Paper

|D

iscussionP

aper|

Discussion

Paper

|D

iscussionP

aper|

introduced to simulate reservoir operations (Van Beek et al., 2011). The outflow fromlakes and reservoir provides the input hydrograph for the downstream river sectionswhile lakes and reservoirs themselves are fed by stream flow where they intersect theriver network.

PCR-GLOBWB includes two options to represent the fraction freshwater. In case it5

is fixed, the areas of lakes, reservoirs and floodplains are kept constant and floodplainsand through flow wetlands (i.e. Niger inner delta) are treated as regular river stretches(albeit with an area and a resistance in terms of Manning’s n that is larger than thoseof the channel proper). If variable, the option used in this study, both river stretchesand water bodies change their area during routing. Discharge volumes in excess of10

the maximum channel storage flood the adjacent land area and pond an area with arelative elevation over the riverbed equivalent to the volume to be stored, which is de-rived from a cumulative distribution of relative elevations based on a 1×1 km resolution(Fig. 7). Over the flooded area, this results in a mean flood depth but also an increasedwetted perimeter. The resulting increase in hydraulic resistance (n in Eq. A3) gener-15

ally will lead to longer travel times and larger associated flood volumes. Together withthe required redistribution of flood waters, this leads to increased computation times.Thus, the schematisation of the freshwater surface area includes direct precipitationand evaporation, but does not include increased infiltration, as the soil hydrology iskept at a daily time step for computational efficiency.20

Parameterisation of the flood model requires information to delineate lakes andreservoirs from rivers, which was taken from the GLWD1 dataset (Lehner and Doll,2004), and a drainage direction map, which was based on the DDM30 (Doll and Lehner,2002). Channel dimensions were derived from hydraulic relationships between bank-full discharge and channel geometry (Allen et al., 1994) and regionalised on the ba-25

sis of the relationship between global climate indicators and the observed bank-fulldischarge for 296 stations of the RivDis dataset (http://www.rivdis.sr.unh.edu/). Chan-nel length was derived directly from the average cell length multiplied by a tortuosityfactor of 1.3. The channel gradient was derived from the 1×1 km elevations from

9642





http://www.rivdis.sr.unh.edu/

HESSD9, 9611–9659, 2012


assessments


Title Page



Tables Figures

J I

J I

Back Close

Full Screen / Esc



Discussion

Paper

|D

iscussionP

aper|

Discussion

Paper

|D

iscussionP

aper|

the Hydro1k (Verdin, 2011) associated with the perennial streams from the VMAPdatabase (http://geoengine.nima.mil/). The same elevation dataset was used to derivethe subgrid cumulative elevation distribution. Starting from a subcatchment distributionwithin each 0.5◦ cell area centred on the perennial streams, the relative height abovethe floodplain of each 1 km cell was calculated. For each cell, the aggregated subgrid5

distribution was then described by selected percentiles (0.01, 0.05 and 0.1 through1.0 by increments of 0.1) and represented by a smooth continuous function followingthe approach of Kavetski and Kuczera (2007). In the absence of detailed informationon Manning’s n, characteristic values of 0.04 and 0.10 were assigned to channel andfloodplains respectively.10

Acknowledgements. We acknowledge the Directorate General of International Cooperationsunder the Ministry of Foreign affairs, The Netherlands for their funding of this research. Further-more, part of this research was funded under a VENI grant from the Netherlands Organisationfor Scientific Research (NWO). NWO is also greatly acknowledged for this support.

References15

Aerts, J. C. J. H., Sprong, T., and Bannink, B. A.: Aandacht voor veiligheid, Leven met Water,Klimaat voor Ruimte, DG Water, Amsterdam, The Netherlands, 2008.

Allen, P. M., Arnold, J. C., and Byars, B. W.: Downstream Channel Geometry for usein planning-level models, J. Am. Water Resour. As., 30, 663–671, doi:10.1111/j.1752-1688.1994.tb03321.x, 1994.20

Balsamo, G., Bousetta, P., Ferranti, L., and Lopez, P.: Evaluation of ERA-Interim and ERA-Interim-GPCP-rescaled precipitation over the USA, ECMWF, Reading, UK, available at: http://www.ecmwf.int/publications/library/do/references/list/782009 (last access: August 2012),2010.

Biemans, H., Hutjes, R. W. A., Kabat, P., Strengers, B. J., Gerten, D., and Rost, S.: Effects of25

Precipitation Uncertainty on Discharge Calculations for Main River Basins, J. Hydrometeo-rol., 10, 1011–1025, 2009.

Bierkens, M. F. P. and van Beek, L. P. H.: Seasonal Predictability of European Discharge: NAOand Hydrological Response Time, J. Hydrometeorol., 10, 953–968, 2009.

9643





http://geoengine.nima.mil/

http://dx.doi.org/10.1111/j.1752-1688.1994.tb03321.x



http://www.ecmwf.int/publications/library/do/references/list/782009



HESSD9, 9611–9659, 2012


assessments


Title Page



Tables Figures

J I

J I

Back Close

Full Screen / Esc



Discussion

Paper

|D

iscussionP

aper|

Discussion

Paper

|D

iscussionP

aper|

Bondeau, A., Smith, P. C., Zaehle, S., Schaphoff, S., Lucht, W., Cramer, W., Gerten, D., Lotze-Campen, H., Muller, C., Reichstein, M., and Smith, B.: Modelling the role of agriculturefor the 20th century global terrestrial carbon balance, Global Change Biol., 13, 679–706,doi:10.1111/j.1365-2486.2006.01305.x, 2007.

Bouwman, A. F., Kram, T., and Klein Goldewijk, K.: Integrated modelling of global environmental5

change. An overview of IMAGE 2.4, The Netherlands Environmental Assessment Agency,2006.

Brakenridge, G. R., Anderson, E., Nghiem, S. V., Caquard, S., and Shabaneh, T.: Flood warn-ings, flood disaster assessments, and flood hazard reduction: the roles of orbital remotesensing, in: Proceedings of the 30th International Symposium on Remote Sensing of the10

Environment, Honolulu, Hawai’i, 2003.De Bruijn, K. M.: Resilience and flood risk management; a systems approach applied to lowland

rivers, PhD thesis, Delft, University of Technology, Delft, The Netherlands, 2005.Carsell, K. M., Pingel, N. D., and Ford, D. T.: Quantifying the Benefit of a Flood Warning System,

Nat. Hazards Review, 5,, 131–140, doi:10.1061/(ASCE)1527-6988(2004)5:3(131), 2004.15

Chow, V.-T.: Open channel hydraulics, McGraw-Hill Book Company, New York, 1988.CIESIN and CIAT: Global Rural-Urban Mapping Project (GRUMP). Center for International

Earth Science Information Network (CIESIN), Columbia University and Centro Internacionalde Agricultura Tropical (CIAT), available at: http://sedac.ciesin.columbia.edu/gpw (last ac-cess: August 2012), 2009.20

Decharme, B., Douville, H., Prigent, C., Papa, F., and Aires, F.: A new river flooding scheme forglobal climate applications: Off-line evaluation over South America, J. Geophys. Res., 113,D11110, doi:200810.1029/2007JD009376, 2008.

Doll, P. and Lehner, B.: Validation of a new global 30-min drainage direction map, J. Hydrol.,258, 214–231, doi:10.1016/S0022-1694(01)00565-0, 2002.25

Ferreira, S., Hamilton, K., and Vincent, J. R.: Nature, Socioeconomics and Adaptation to NaturalDisasters: New Evidence from Floods, SSRN eLibrary, available at: http://papers.ssrn.com/sol3/papers.cfm?abstract id=1876300 (last access; 22 May 2012), 2011.

Field, C. B., Barros, V., Stocker, T. F., Qin, D., Dokken, D., Ebi, K. L., Mastrandrea, M. D., Mach,K. J., Plattner, G.-K., Allen, S. K., Tignor, M., and Midgley, P. M.: Summary for policymakers,30

in Intergovernmental Panel on Climate Change – Special Report on Managing the Risks ofExtreme Events and Disasters to Advance Climate Change Adaptation, Cambridge Univer-sity Press, Cambridge and New York, 2011.

9644





http://dx.doi.org/10.1111/j.1365-2486.2006.01305.x

http://dx.doi.org/10.1061/(ASCE)1527-6988(2004)5:3(131)

http://sedac.ciesin.columbia.edu/gpw

http://dx.doi.org/10.1016/S0022-1694(01)00565-0

http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1876300



HESSD9, 9611–9659, 2012


assessments


Title Page



Tables Figures

J I

J I

Back Close

Full Screen / Esc



Discussion

Paper

|D

iscussionP

aper|

Discussion

Paper

|D

iscussionP

aper|

Green, C. H.: Coastal cities: assets at risk and depth-damage curves, Report, prepared for theOECD, Middlesex University, Middlesex, 2010.

Green, C. H., Viavattene, C., and Thompson, P.: Guidance for assessing flood losses, CONHAZreport, Flood Hazard Research Centre, Middlesex University, Middlesex, 2011.

Gupta, H. V., Wagener, T., and Liu, Y.: Reconciling theory with observations: ele-5

ments of a diagnostic approach to model evaluation, Hydrol. Process., 22, 3802–3813,doi:10.1002/hyp.6989, 2008.

Haddeland, I., Clark, D. B., Franssen, W., Ludwig, F., Voß, F., Arnell, N. W., Bertrand, N., Best,M., Folwell, S., Gerten, D., Gomes, S., Gosling, S. N., Hagemann, S., Hanasaki, N., Harding,R., Heinke, J., Kabat, P., Koirala, S., Oki, T., Polcher, J., Stacke, T., Viterbo, P., Weedon, G.10

P., and Yeh, P.: Multimodel Estimate of the Global Terrestrial Water Balance: Setup and FirstResults, J. Hydrometeorol., 12, 869–884, doi:10.1175/2011JHM1324.1, 2011.

Hill, H.: Intra-country regional disparities, in Proceedings of the Second Asian DevelopmentConference, Asian Development Bank, Singapore, 2000.

Huffman, G. J., Adler, R. F., Bolvin, D. T., and Gu, G.: Improving the global precipitation record:15

GPCP Version 2.1, Geophys. Res. Lett., 36, L17808, doi:200910.1029/2009GL040000,2009.

IPCC: IPCC special report – Emissions scenarios – IPCC working group III, IPCC, availableat: (last access: 3 May 2012), 2000.

Jha, A. K., Bloch, R., and Lamond, J.: Cities and flooding: a guide to integrated urban flood risk20

management for the 21st century, The World Bank, Washington D. C., 2012.Jongman, B., Ward, P. J., and Aerts, J. C. J. H.: Global exposure to river

and coastal flooding: Long term trends and changes, Global Environ. Change,doi:10.1016/j.gloenvcha.2012.07.004, in press, 2012a.

Jongman, B., Kreibich, H., Bates, P. D., Barredo, J. I., Gericke, A., Apel, H., Neal, J., Aerts, J. C.25

J. H., and Ward, P. J.: Comparative flood damage model assessment: Towards a Europeanapproach, EGU2012–7663, EGU, Vienna, Austria, 2012b.

Jonkman, S. N.: Loss of life estimation in flood risk assessment, Theory and applications, PhDthesis, Delft, University of Technology, Delft, The Netherlands, 2007.

Kaufmann, D., Kraay, A., and Mastruzzi, M.: The Worldwide Governance Indicators (WGI)30

project, available at: http://info.worldbank.org/governance/wgi/index.asp, last access: Au-gust 2012, The World Bank, Washington D. C., 2011.

9645





http://dx.doi.org/10.1002/hyp.6989

http://dx.doi.org/10.1175/2011JHM1324.1

http://dx.doi.org/10.1016/j.gloenvcha.2012.07.004

http://info.worldbank.org/governance/wgi/index.asp

HESSD9, 9611–9659, 2012


assessments


Title Page



Tables Figures

J I

J I

Back Close

Full Screen / Esc



Discussion

Paper

|D

iscussionP

aper|

Discussion

Paper

|D

iscussionP

aper|

Kavetski, D. and Kuczera, G.: Model smoothing strategies to remove microscale discontinu-ities and spurious secondary optima in objective functions in hydrological calibration, WaterResour. Res., 43, W03411, doi:200710.1029/2006WR005195, 2007.

Klein Goldewijk, K., Beusen, A., van Drecht, G. and de Vos, M.: The HYDE 3.1 spatially explicitdatabase of human-induced global land-use change over the past 12,000 years, Global Ecol.5

Biogeogr., 20, 73–86, doi:10.1111/j.1466-8238.2010.00587.x, 2011.Klijn, F., Baan, P. J., Bruijn, K. M. de, Kwadijk, J. C., and Buren, R. van: Overstromingsrisico’s in

Nederland in Een Veranderend Klimaat Verwachtingen, Schattingen En Berekeningen VoorHet Project Nederland Later, WL Delft Hydraulics, Delft, 2007.

Kummu, M., de Moel, H., Ward, P. J., and Varis, O.: How Close Do We Live to Water?10

A Global Analysis of Population Distance to Freshwater Bodies, PLoS ONE, 6, e20578,doi:10.1371/journal.pone.0020578, 2011.

Lehner, B. and Doll, P.: Development and validation of a global database of lakes, reservoirsand wetlands, J. Hydrol., 296, 1–22, doi:10.1016/j.jhydrol.2004.03.028, 2004.

Letourneau, A., Verburg, P. H., and Stehfest, E.: A land-use systems approach to represent15

land-use dynamics at continental and global scales, Environ. Model. Softw., 33, 61–79,2012.

Meijer, J., Hilderink, H., and Lucas, P.: Exploring data and methods used for global modellingof the urban population and extent, The Netherlands Environmental Assessment Agency,Bilthoven, The Netherlands, 2006.20

Merz, B., Kreibich, H., Schwarze, R., and Thieken, A.: Review article “Assessment of economicflood damage”, Nat. Hazards Earth Syst. Sci., 10, 1697–1724, doi:10.5194/nhess-10-1697-2010, 2010.

Messner, F., Penning Rowsell, E. C., Green, C., Meyer, V., Tunstall, S. M., and Van der Veen,A.: Evaluating flood damages: guidance and recommendations on principles and methods,25

FLOODsite, Wallingford, UK, 2007.Monirul Qader Mirza, M.: Three Recent Extreme Floods in Bangladesh: A Hydro-Meteorological

Analysis, Nat. Hazards, 28, 35–64, doi:10.1023/A:1021169731325, 2003.Munich, Re: Topics Geo, natural catastrophes 2009: analyses, assessments, positions, Munich

Reinsurance Company, Munich, Germany, 2010.30

Neal, J., Schumann, G., Hall, A., and Bates, P.: A simple model for simulating river hydraulicsover large and data sparse areas, Water Resour. Res., in preparation, 2012.

9646





http://dx.doi.org/10.1111/j.1466-8238.2010.00587.x

http://dx.doi.org/10.1371/journal.pone.0020578

http://dx.doi.org/10.1016/j.jhydrol.2004.03.028

http://dx.doi.org/10.5194/nhess-10-1697-2010



http://dx.doi.org/10.1023/A:1021169731325

HESSD9, 9611–9659, 2012


assessments


Title Page



Tables Figures

J I

J I

Back Close

Full Screen / Esc



Discussion

Paper

|D

iscussionP

aper|

Discussion

Paper

|D

iscussionP

aper|

New, M., Lister, D., Hulme, M., and Makin, I.: A high-resolution data set of surface climate overglobal land areas, Clim. Res., 21, 1–25, 2002.

Pappenberger, F., Dutra, E., Wetterhall, F., and Cloke, H.: Deriving global flood hazard maps offluvial floods through a physical model cascade, Hydrol. Earth Syst. Sci. Discuss., 9, 6615–6647, doi:10.5194/hessd-9-6615-2012, 2012.5

PBL: Towards a Global Integrated Sustainability Model, GISMO 1.0 status report, PBL Nether-lands Environmental Assessment Agency, 2008.

Peduzzi, P., Dao, H., Herold, C., and Mouton, F.: Assessing global exposure and vulnerabilitytowards natural hazards: the Disaster Risk Index, Nat. Hazards Earth Syst. Sci., 9, 1149–1159, doi:10.5194/nhess-9-1149-2009, 2009.10

Prigent, C., Papa, F., Aires, F., Rossow, W. B. and Matthews, E.: Global inundation dynamicsinferred from multiple satellite observations, 1993–2000, J. Geophys. Res., 112, D12107,doi:10.1029/2006JD007847, 2007.

Rost, S., Gerten, D., Bondeau, A., Lucht, W., Rohwer, J. and Schaphoff, S.: Agricultural greenand blue water consumption and its influence on the global water system, Water Resour.15

Res., available at: http://www.agu.org/pubs/crossref/2008/2007WR006331.shtml (last ac-cess: 8 October 2009), 2008.

Ruff, T. W. and Neelin, J. D.: Long tails in regional surface temperature probability distribu-tions with implications for extremes under global warming, Geophys. Res. Lett., 39, L05804,doi:201210.1029/2011GL050610, 2012.20

Schneider, A., Friedl, M. A., and Potere, D.: A new map of global urban extent from MODISsatellite data, Environ. Res. Lett., 4, 044003, doi:10.1088/1748-9326/4/4/044003, 2009.

Seto, K. C., Fragkias, M., Guneralp, B., and Reilly, M. K.: A Meta-Analysis of Global Urban LandExpansion, PLoS ONE, 6, e23777, doi:10.1371/journal.pone.0023777, 2011.

Seto, K. C. and Shepherd, J. M.: Global urban land-use trends and climate impacts, Current25

Opinion in Environmental Sustainability, 1, 89–95, doi:10.1016/j.cosust.2009.07.012, 2009.Smith, D. I.: Flood damage estimation – A review of urban stage-damage curves and loss

functions, Water SA (online), 20, 231–238, 1994.Sperna Weiland, F. C.: Hydrological impacts of climate change?: interpretation of uncertainties

introduced by global models of climate and hydrology, Utrecht Studies in Earth Sciences,30

006, available at: http://igitur-archive.library.uu.nl/dissertations/2011-1130-200336/UUindex.html (last access: 12 June 2012), 2011.

9647





http://dx.doi.org/10.5194/hessd-9-6615-2012


http://dx.doi.org/10.1029/2006JD007847

http://www.agu.org/pubs/crossref/2008/2007WR006331.shtml

http://dx.doi.org/10.1088/1748-9326/4/4/044003

http://dx.doi.org/10.1371/journal.pone.0023777

http://dx.doi.org/10.1016/j.cosust.2009.07.012

http://igitur-archive.library.uu.nl/dissertations/2011-1130-200336/UUindex.html



HESSD9, 9611–9659, 2012


assessments


Title Page



Tables Figures

J I

J I

Back Close

Full Screen / Esc



Discussion

Paper

|D

iscussionP

aper|

Discussion

Paper

|D

iscussionP

aper|

Sperna Weiland, F. C., van Beek, L. P. H., Kwadijk, J. C. J., and Bierkens, M. F. P.: The abilityof a GCM-forced hydrological model to reproduce global discharge variability, Hydrol. EarthSyst. Sci., 14, 1595–1621, doi:10.5194/hess-14-1595-2010, 2010.

Strahler, A. N.: Handbook of AApplied Hydrology, edited by: Chow, V. T., 4–39 4–76, McGraw-Hill, New York, 1964.5

Tapsell, S. M., Penning-Rowsell, E. C., Tunstall, S. M. and Wilson, T. L.: Vulnerability toFlooding: Health and Social Dimensions, Phil. Trans. R. Soc. Lond. A, 360, 1511–1525,doi:10.1098/rsta.2002.1013, 2002.

UNISDR: Hyogo Framework for Action 2005–2015: Building the Resilience of Nations and Com-munities to Disasters, United Nations, Kobe, Japan, 2005.10

UNISDR: Global Assessment Report on Disaster Risk Reduction. Risk and Poverty in aChanging Climate, United Nations International Strategy for Disaster Reduction Secretariat,Geneva, 2009.

UNISDR: Global Assessment Report on Disaster Risk Reduction. Revealing Risk, Redefin-ing Development, United Nations International Strategy for Disaster Reduction Secretariat,15

Geneva, 2011.Uppala, S. M., Kallberg, P. W., Simmons, A. J., Andrae, U., Bechtold, V. D., Fiorino, M., Gib-

son, J. K., Haseler, J., Hernandez, A., and Kelly, G. A.: The ERA-40 re-analysis, Q. J. Roy.Meteorol. Soc., 131, 2961–3012, doi:10.1256/qj.04.176, 2005.

Van Beek, L. P. H. and Bierkens, M. F. P.: The global hydrological model PCR-GLOBWB:20

conceptualization, parameterization and verification, Dept. of Physical Geography, UtrechtUniversity, Utrecht, available at: http://vanbeek.geo.uu.nl/suppinfo/vanbeekbierkens2009.pdf(last access: August 2012), 2009.

Van Beek, L. P. H., Wada, Y., and Bierkens, M. F. P.: Global monthly water stress: I. Water bal-ance and water availability, Water Resour. Res., 47, W07517, doi:10.1029/2010WR009791,25

2011.Verburg, P. H., Neumann, K., and Nol, L.: Challenges in using land use and land cover

data for global change studies, Global Change Biol., 17, 974–989, doi:10.1111/j.1365-2486.2010.02307.x, 2011.

Verdin, K. L.: ISLSCP II HYDRO1k Elevation-derived Products, in ISLSCP Initiative II Collec-30

tion. Data set, Oak Ridge National Laboratory Distributed Active Archive Center, Oak Ridge,Tennessee, USA, available at: http://daac.ornl.gov/ (last access: August 2012), 2011.

9648





http://dx.doi.org/10.5194/hess-14-1595-2010

http://dx.doi.org/10.1098/rsta.2002.1013

http://dx.doi.org/10.1256/qj.04.176

http://vanbeek.geo.uu.nl/suppinfo/vanbeekbierkens2009.pdf

http://dx.doi.org/10.1029/2010WR009791

http://dx.doi.org/10.1111/j.1365-2486.2010.02307.x

http://dx.doi.org/10.1111/j.1365-2486.2010.02307.x

http://dx.doi.org/10.1111/j.1365-2486.2010.02307.x

http://daac.ornl.gov/

HESSD9, 9611–9659, 2012


assessments


Title Page



Tables Figures

J I

J I

Back Close

Full Screen / Esc



Discussion

Paper

|D

iscussionP

aper|

Discussion

Paper

|D

iscussionP

aper|

Verkade, J. S. and Werner, M. G. F.: Estimating the benefits of single value and probabilityforecasting for flood warning, Hydrol. Earth Syst. Sci., 15, 3751–3765, doi:10.5194/hess-15-3751-2011, 2011.

van Vuuren, D. P., Lucas, P. L., and Hilderink, H.: Downscaling drivers of global environmen-tal change: Enabling use of global SRES scenarios at the national and grid levels, Global5

Environ. Change, 17, 114–130, doi:10.1016/j.gloenvcha.2006.04.004, 2007.Weedon, G. P., Gomes, S., Viterbo, P., Shuttleworth, W. J., Blyth, E., Osterle, H., Adam, J. C.,

Bellouin, N., Boucher, O., and Best, M.: Creation of the WATCH Forcing Data and Its Useto Assess Global and Regional Reference Crop Evaporation over Land during the TwentiethCentury, J. Hydrometeorol., 12, 823–848, doi:10.1175/2011JHM1369.1, 2011.10

Weerts, A. H., Winsemius, H. C., and Verkade, J. S.: Estimation of predictive hydrological un-certainty using quantile regression: examples from the National Flood Forecasting System(England and Wales), Hydrol. Earth Syst. Sci., 15, 255–265, doi:10.5194/hess-15-255-2011,2011.

Werner, M. G. F., Schellekens, J., and Kwadijk, J. C.: Flood Early Warning Systems for Hydro-15

logical (sub) catchments, in: Encyclopedia of Hydrological Sciences, 1, 349–364, John Wiley& Sons, 2005.

Westerberg, I. K., Guerrero, J.-L., Younger, P. M., Beven, K. J., Seibert, J., Halldin, S., Freer,J. E., and Xu, C.-Y.: Calibration of hydrological models using flow-duration curves, Hydrol.Earth Syst. Sci., 15, 2205–2227, doi:10.5194/hess-15-2205-2011, 2011.20

Winsemius, H. C., Schaefli, B., Montanari, A. and Savenije, H. H. G.: On the calibration of hy-drological models in ungauged basins: A framework for integrating hard and soft hydrologicalinformation, Water Resour. Res., 45, W12422, doi:200910.1029/2009WR007706, 2009.

World Bank: Economics of Adaptation to Climate Change. Bangladesh, World Bank, Washing-ton D. C., 2010.25

World Bank: World development indicators, available at: http://databank.worldbank.org/ddp/home.do, World Bank, Washington D. C., 2012.

Yamazaki, D., Kanae, S., Kim, H., and Oki, T.: A physically based description of floodplaininundation dynamics in a global river routing model, Water Resour. Res., 47, W04501,doi:201110.1029/2010WR009726, 2011.30

9649








http://dx.doi.org/10.1016/j.gloenvcha.2006.04.004

http://dx.doi.org/10.1175/2011JHM1369.1



http://databank.worldbank.org/ddp/home.do



HESSD9, 9611–9659, 2012


assessments


Title Page



Tables Figures

J I

J I

Back Close

Full Screen / Esc



Discussion

Paper

|D

iscussionP

aper|

Discussion

Paper

|D

iscussionP

aper|

Table 1. Required model characteristics.

Characteristic Value Justification

Forcing > 30−yr daily dataset, comprised of ob-servations, in which the day-to-day variabil-ity and auto-correlation is as much as possi-ble preserved. Reanalysis records, in whichprecipitation observations are assimilated,can also be used.

In particular, precipitation should be vol-umetrically as accurate as possible, butshould also contain the temporal character-istics of rainfall, because flood genesis istypically dependent on multi-day rainfall ac-cumulations.

Model time step (Sub)-Daily Runoff generation is a highly non-linear pro-cess, and should be resolved at sufficientlyshort time scales. Furthermore flood prop-agation over typical grid cell sizes used inglobal hydrology occurs at daily or evensub-daily time scale

Potential evaporation scheme Radiation based approach Haddeland et al. (2011) showed that anon-radiation based approach may result inoverestimation of potential and thus actualevaporation during storm (and hence flood)periods.

Runoff scheme Infiltration excess as non-linear function ofsoil moisture

A non-linear relation with soil moisture pro-vides the most realistic runoff generationin time and therefore the best hydrographshape

Routing Dynamic routing with sub-grid variableoverbank elevation

A dynamic routing, which differentiates riverflow from overbank flow, is required to sim-ulate sub-grid flood extent and depth. Thiscomponent can be a separate model, forcedby the outputs of a global hydrologicalmodel.

9650





HESSD9, 9611–9659, 2012


assessments


Title Page



Tables Figures

J I

J I

Back Close

Full Screen / Esc



Discussion

Paper

|D

iscussionP

aper|

Discussion

Paper

|D

iscussionP

aper|

Table 2. Expected annual damage (land use and population method approach) for theBangladesh case study for the reference scenario and for the future scenarios (2050) withhazard change only (ECHAM, HADGEM), exposure change only, and hazard and exposurechange combined.

Land use method ($ PPP millions) Population method ($ PPP millions)

Exposure scenario Ref. asset value 2050 asset value Ref. GDP 2050 GDPClimate scenarioReference climate 740 8171 2183 14 7912050 (ECHAM) 2017 22 128 6996 62 7822050 (HADGEM) 2701 30 061 9281 46 327

9651





HESSD9, 9611–9659, 2012


assessments


Title Page



Tables Figures

J I

J I

Back Close

Full Screen / Esc



Discussion

Paper

|D

iscussionP

aper|

Discussion

Paper

|D

iscussionP

aper|

Table 3. Comparison of observed and simulated damages in 1998 and 2004. Note: for the1998 flood we used the modelled results for the 30 yr return period inundation when in realitythe return period was in the order of 30–70 yr; and for the 2004 flood we used the modelledresults for the 15 yr return period inundation when in reality the return period was in the orderof 20 yr. All values are in $ PPP2010.

$ PPP (2010 values)

1998 2004EM-DAT 12 313 6695World Bank (2010) 6093 5660Modelled expected annualdamage (land use approach)

4635 3415

Modelled expected annualdamage (population method)

16 837 11 995

9652





HESSD9, 9611–9659, 2012


assessments


Title Page



Tables Figures

J I

J I

Back Close

Full Screen / Esc



Discussion

Paper

|D

iscussionP

aper|

Discussion

Paper

|D

iscussionP

aper|

Global forcing climate

Prec. (x,t)Temp. (x,t)

Potential evap. (x,t)

Global runoff

Modelled specific river discharge (x,t)

Global inundation

Global inundation extent and depth (x,t)

Local hazard

Probability densityof flood hazard

Local exposure

Socio-economicindicators

Global exposure

Socio-economicindicators from global

assessments / databases

downscalingmethods

downscalingmethods

Global routingmodels

Global hydrologicalmodels

Global forcingdatasets

Framework for Global Flood Risk assessment

Global scale

Local scaleFlood Risk

Local vulnerability

Resilience, Adaptationcapacity

Fig. 1. Schematic of framework for global flood risk assessment.

9653





HESSD9, 9611–9659, 2012


assessments


Title Page



Tables Figures

J I

J I

Back Close

Full Screen / Esc



Discussion

Paper

|D

iscussionP

aper|

Discussion

Paper

|D

iscussionP

aper|

Fig. 2. An illustration of the sequential leveling of river water levels with the surrounding con-nected pixels over a part of Bangladesh on a 1×1 km elevation grid. The examples shown hereare computed using the once in 30 yr flood from the ERA40/CRU reference scenario; assumingflooding from stream order 6 and higher; a levee height, conform a recurrence interval of 0 yr(i.e. the total volume from DynRout is considered): (a) leveling with 10 cm of water, (b) 20 cm ofwater, (c) 2 m of water, some areas have stopped filling, (d) 10 m of water.

9654





HESSD9, 9611–9659, 2012


assessments


Title Page



Tables Figures

J I

J I

Back Close

Full Screen / Esc



Discussion

Paper

|D

iscussionP

aper|

Discussion

Paper

|D

iscussionP

aper|

5

0

21

0

0

0-1

1-2

2-3

3-4

4-5

>5

Inundation depth(m)

Expected annual damage($ million / year)

Expected annual affected GDP($ million / year)

Fig. 3. From left to right: the 30-yr flood, downscaled to Bangladesh; the expected value ofannual damage (land use method); and the expected value of annual affected GDP (populationmethod), based on the reference climate (1961–1990) and the current population, land use,and GDP

9655





HESSD9, 9611–9659, 2012


assessments


Title Page



Tables Figures

J I

J I

Back Close

Full Screen / Esc



Discussion

Paper

|D

iscussionP

aper|

Discussion

Paper

|D

iscussionP

aper|

10 10

20 20

30 30

70

70

80

80

90

90

100

100

110

110

120

120

Fig. 4. Once in 30-yr inundation according to GLOFRIS model cascade, overlaid on the Dart-mouth Flood Observatory maximum inundation extent.

9656





HESSD9, 9611–9659, 2012


assessments


Title Page



Tables Figures

J I

J I

Back Close

Full Screen / Esc



Discussion

Paper

|D

iscussionP

aper|

Discussion

Paper

|D

iscussionP

aper|

0

0-1

1-2

2-3

3-4

4-5

>5

Inundation depth(m)

Inundation extent (DFO)Not inundated

Inundated

Fig. 5. Validation of flood hazard map over Bangladesh.

9657





HESSD9, 9611–9659, 2012


assessments


Title Page



Tables Figures

J I

J I

Back Close

Full Screen / Esc



Discussion

Paper

|D

iscussionP

aper|

Discussion

Paper

|D

iscussionP

aper|

Fig. 6. The sensitivity of the downscaling to the 2 choices: (1) from which Strahler streamorder do river floods occur?; and (2) at which minimum return period do floods start to causedamage? Each plot shows a top-view and zonal averages in longitudinal and latitudinal directionof inundation levels [m] at a 30-yr return period. Each subfigure shows the results with differentassumptions on the river size, and return periods at which river floods start to play a role.

9658





HESSD9, 9611–9659, 2012


assessments


Title Page



Tables Figures

J I

J I

Back Close

Full Screen / Esc



Discussion

Paper

|D

iscussionP

aper|

Discussion

Paper

|D

iscussionP

aper|

Area fractionFlooded Area fraction =

Dynamic �oodplain extent

Discharge volume in excess of bank-full

Bank-full

Cumulative frequency relative elevation

Relative elevation

Fig. 7. Schematic of determining the static floodplain area and the dynamic floodplain extent(fraction flooded area) within a 0.5×0.5◦ cell based on bank-full discharge surface water level,actual surface water level and a 1×1 km digital elevation model. The discharge volume inexcess of bank-full, as shown in the top figure, is mapped onto the cumulative distribution offloodplain elevation, as shown in the bottom figure.

9659





Documents

A framework for global river flood risk assessments